PLC Splitter Guide: Split Ratios, Insertion Loss Charts, and Placement in the ODN

A PLC splitter (planar lightwave circuit splitter) is the passive device that divides one optical signal into many in a PON — the "P" for passive in passive optical network. One fiber from the OLT feeds the splitter; the splitter fans it out to 2, 4, 8, 16, 32 or 64 subscribers, with no power and no electronics. The single number that governs how you design around it is insertion loss: every split costs light, and the split ratio you choose has to fit inside the PON's power budget. This guide gives the real insertion-loss figures by split ratio, shows how to spend them against a GPON budget, and covers where the splitter sits in the ODN.

Key takeaways

• Insertion loss rises with split ratio — roughly 3–3.5 dB per doubling. A 1×32 splitter costs about 16.3 dB, a 1×64 about 19.8 dB.
• The split ratio is a power-budget decision, not a "how many subscribers" decision. Pick the largest split the budget allows after fiber, connector and margin losses.
• Loss uniformity and PDL matter for the worst-case port — the budget must close on the weakest output, not the average.
• PLC beats FBT above 1×8 — uniform loss across all ports and the full 1260–1650 nm band, where fused splitters drift.
• Placement (centralized vs cascaded) trades fiber count against flexibility; the total split — not the number of stages — is what the budget sees.

PLC splitter insertion loss by split ratio

This is the table the whole design turns on. These are DYS published maximums — the worst-case insertion loss you can budget against, across the full 1260–1650 nm window:

Two things to read out of it. First, the loss is close to the theoretical split loss plus a small excess: an ideal 1×32 split spreads power over 32 ports = 15 dB of pure division, and the ≤16.3 dB figure means only ~1.3 dB of excess loss on top. Second, uniformity and PDL are separate budget items. Uniformity is the spread between the best and worst output port — on a 1×32, up to 1.5 dB — and your budget has to close on the worst port, so the number that matters for the last subscriber is closer to insertion loss plus uniformity. Additional device specs: return loss UPC ≥50 dB / APC ≥60 dB, directivity ≥55 dB, on 9/125 µm SMF‑28e fiber, rated −40 to +85 °C.

Why loss grows with the split — and why it's not linear

Splitting is division of optical power. Send 1 mW into a 1×2 and each output gets half — that alone is 3 dB, before any device imperfection. Every further doubling costs another ~3 dB of pure split loss:

The takeaway for a planner: the insertion-loss penalty for going from 1×16 to 1×32 is only ~3 dB, but it doubles your subscriber reach per feeder fiber. That trade — 3 dB for 2× the customers — is the core economics of PON, and it is why high-split PONs are attractive right up to the point the budget runs out.

Spending the loss: a GPON power-budget worked example

A PON link "closes" when the total loss from OLT to ONU stays under the optical budget of the transceiver class. Standard GPON classes (ITU‑T G.984) give roughly 28 dB (Class B+) or 32 dB (Class C+) of budget. Here is how a 1×32 build spends a Class B+ budget:

That closes inside 28 dB — but only just, and only because the fiber run is modest. Push to 1×64 (19.8 + 2.0 = 21.8 dB on the worst port) and the same link needs ~30.8 dB — over Class B+, so you would move to Class C+ optics or shorten the fiber. This is the whole point of the insertion-loss table: the split ratio is chosen by working backwards from the budget, not by how many homes you wish to serve. See the fiber size and loss references for the fiber-side figures.

Placement in the ODN: centralized vs cascaded

The optical distribution network (ODN) is everything between the OLT and the subscribers, and the splitter can sit in one place or be split across stages:

The key insight: the budget sees the total split, and the two are within about 1 dB — a cascade costs marginally more because the second device adds another dose of excess loss and its own uniformity, so a 1×4 + 1×8 cascade lands near ~18.8 dB against a single 1×32's ~17.8 dB. That gap rarely decides a link, so the choice is a fiber-and-flexibility decision, not a loss decision. Centralized keeps all splitters in one accessible place; cascaded pushes the final split closer to subscribers, which cuts drop-fiber count and suits phased roll-outs where take-up is uncertain. For the last-drop cable that follows the splitter, see the FTTH deployment guides.

PLC vs FBT: why planar wins above 1×8

Two splitter technologies exist. FBT (fused biconical taper) fuses fibers together — cheap at 1×2 and 1×4, but loss climbs unevenly as ports increase and it is wavelength-sensitive. PLC etches the splitter as a waveguide circuit on a chip, which gives uniform loss across every port and across the full 1260–1650 nm band. For any PON above 1×8 — and any network carrying 1310/1490/1550 nm together — PLC is the standard choice, and it is what the insertion-loss table above describes. FBT survives only at low split ratios where its cost edge still matters.

PLC splitters ship in several packages for where they mount: mini for splice trays and closures, ABS-box and cassette for distribution boxes, and rack-mount for the central office or hub. The optical performance is the same across packages — the choice is mechanical, driven by where in the ODN the splitter lands. DYS supplies PLC splitters from 1×2 to 1×64 across these formats; the DYS product range lists the configurations.

Frequently asked questions

What is a PLC splitter?

A planar lightwave circuit splitter is a passive optical component that divides one input signal into multiple outputs — from 1×2 up to 1×64 — using a waveguide circuit etched on a chip. It needs no power and works across the full 1260–1650 nm band, which makes it the standard power splitter in FTTH/PON access networks.

What is the insertion loss of a 1×32 splitter?

A 1×32 PLC splitter has a maximum insertion loss of about 16.3 dB — roughly 15 dB of theoretical split loss (dividing power over 32 ports) plus around 1.3 dB of device excess loss. For worst-case budgeting, add the loss uniformity (up to 1.5 dB), giving about 17.8 dB on the weakest output port.

How much loss does each split ratio add?

Roughly 3–3.5 dB per doubling of ports: 1×2 ≤3.9 dB, 1×4 ≤7.1 dB, 1×8 ≤10.3 dB, 1×16 ≤13.3 dB, 1×32 ≤16.3 dB, 1×64 ≤19.8 dB (DYS maximums). Each doubling roughly halves the power per port, which is the 3 dB, plus a small rise in excess loss.

What split ratio should I use for GPON?

Work backwards from the power budget. A standard GPON Class B+ link has about 28 dB of budget; after fiber, connectors and margin, 1×32 typically closes on modest runs while 1×64 usually needs Class C+ optics or shorter fiber. Choose the largest split that closes on the worst-case port within your budget.

What is the difference between a PLC and an FBT splitter?

PLC (planar lightwave circuit) etches the splitter on a chip, giving uniform loss across all ports and the full 1260–1650 nm band. FBT (fused biconical taper) fuses fibers and is cheaper at low ratios but loses uniformity and wavelength stability as ports grow. PLC is preferred for 1×8 and above and for multi-wavelength PONs.

Does loss uniformity matter, or just insertion loss?

Both. Insertion loss is the nominal per-port loss; uniformity is the spread between the best and worst output. Because a PON link must close on the weakest port, the design figure is insertion loss plus uniformity — on a 1×32 that is up to 16.3 + 1.5 = 17.8 dB. Ignoring uniformity is a common way to under-budget a build.

Where does the splitter go in the network?

In the ODN between the OLT and subscribers — either centralized (one splitter in a hub, simplest to manage) or cascaded (e.g. 1×4 at the cabinet then 1×8 near buildings, which cuts drop-fiber count and suits phased builds). The power budget sees the total split either way, so placement is a fiber-count and flexibility decision, not a loss one.